Check patentability & draft patents in minutes with Patsnap Eureka AI!

Mechanical arm anti-interference motion planning method based on multi-agent reinforcement learning

A multi-agent and reinforcement learning technology, applied in manipulators, program-controlled manipulators, claw arms, etc., can solve problems such as weak anti-interference ability, achieve strong anti-interference ability, improve robustness, and improve anti-interference ability.

Active Publication Date: 2022-02-25
HARBIN INST OF TECH
View PDF7 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of weak anti-interference ability of the mechanical arm reinforcement learning motion planning algorithm based on reinforcement learning, the present invention further proposes a multi-agent reinforcement learning-based mechanical arm anti-interference motion planning method

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Mechanical arm anti-interference motion planning method based on multi-agent reinforcement learning
  • Mechanical arm anti-interference motion planning method based on multi-agent reinforcement learning
  • Mechanical arm anti-interference motion planning method based on multi-agent reinforcement learning

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0053] Specific embodiment one: a kind of anti-jamming motion planning algorithm of the manipulator based on multi-agent reinforcement learning, comprises the following steps:

[0054] Step 1: Discretize the manipulator into a multi-intelligence form based on the joint graph;

[0055] Step 11: Take the joints of the n-degree-of-freedom manipulator as nodes V={1,2,...,n} in the graph, and the links between the joints as edges ε, then the joint graph of the manipulator can be expressed as It is an undirected graph G=(V,ε);

[0056] Step 1 and 2: Each agent can select joint nodes from the joint graph to build its own sub-joint graph, and each agent can only control the joint nodes in its own sub-joint graph;

[0057] Step 13: Design the observation space for each agent. The observation information of each agent consists of three parts. The first part is the state information of each joint node in the agent's own sub-joint graph (such as joint angle, angular velocity, torque inf...

Embodiment

[0083] 1) Experimental tasks

[0084] The training scene for multi-agent manipulator motion planning is a desktop scene built based on the MuJoCo physical simulation engine, such as figure 1 As shown, the task of the manipulator is to move to the target position without colliding with the environment (the target position point is represented by q goal and the forward kinematics model of the manipulator), and the planning is considered successful when the distance between the origin of the coordinate system at the end of the manipulator and the target position is less than 1 cm. The initial configuration q of the training process start and the target configuration q goal are randomly selected within a certain range.

[0085] 2) Multi-agent decomposition of the robotic arm

[0086] The present invention divides the manipulator into three situations: single-agent, double-agent and three-agent based on the joint graph. figure 2 shown. Single agent refers to all the joint no...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a mechanical arm anti-interference motion planning method based on multi-agent reinforcement learning, and belongs to the technical field of robot motion planning and intelligent control. The invention aims to solve the problem of weak anti-interference capability of a mechanical arm neural motion planner based on reinforcement learning. The method comprises the steps: providing a multi-agent decomposition method for a single mechanical arm through establishment of a joint diagram of the mechanical arm and analysis of an association relationship; and the multi-agent SAC reinforcement learning algorithm based on the centralized learning architecture realizes the training of multi-agent reinforcement learning of the mechanical arm. By applying action disturbance, joint locking and observation disturbance in the motion planning process, it is verified that the discretized multi-agent mechanical arm reinforcement learning motion planning method has higher anti-interference capacity compared with a traditional single-agent mechanical arm. The method is applied to the field of motion planning and intelligent control of the mechanical arm.

Description

technical field [0001] The invention relates to a mechanical arm motion planning method based on multi-agent reinforcement learning, and belongs to the technical field of robot motion planning and intelligent control. Background technique [0002] In recent years, with the rapid development of artificial intelligence technology, deep neural networks represented by reinforcement learning have gradually shown their advantages in solving high-dimensional complex problems, providing new ideas for efficient autonomous motion planning of robotic arms. Compared with the traditional motion planning method, thanks to the powerful learning ability of the neural network for the motion planning strategy and the adjustable exploration mechanism in the process of interacting with the environment, it can realize fast online planning in the high-dimensional motion space, thus having Faster response speed and higher execution efficiency. But at the same time, the existing motion planning so...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): B25J9/16B25J17/02B25J18/00
CPCB25J9/1664B25J9/1602B25J9/1679B25J17/0258B25J18/00
Inventor 白成超郭继峰张家维
Owner HARBIN INST OF TECH
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com